K D MOHANA SUNDARAM

@sietk.org

Assistant Professor, Department of Electronics and Communication Engineering
SIDDHARTH INSTITUTE OF ENGINEERING & TECHNOLOGY

K D MOHANA SUNDARAM


               Download Compact PDF  
https://researchid.co/kdmsundaram

RESEARCH, TEACHING, or OTHER INTERESTS

Engineering, Engineering, Engineering, Engineering
7

Scopus Publications

Scopus Publications

  • Lemon fruit classification by transfer learning technique: experimental investigation of convolutional neural network
    K.D. Mohana Sundaram, T. Shankar, N. Sudhakar Reddy
    International Journal of Information and Decision Sciences, 2025
    Before exporting fruits, quality control is extremely important in the fruit industries. The most crucial step in the quality assessment process is to classify the fruit as fresh or spoiled. Convolutional neural network (CNN) is the most recent technology used for classification. Henceforth, in this work, the performance of eight widely used CNNs, namely AlexNet, DenseNet, GoogleNet, Inceptionv-3, MobileNetv-2, ResNet-18, SqueezNet, and VGGNet-19, was evaluated and compared for fruit classification, utilising the Lemon fruit dataset. To classify the lemon fruits into three categories of good-quality, medium-quality, and poor-quality, 1,000 fully connected layers in each CNN were substituted with three fully connected layers. For comparison, all of the CNNs were trained using the Transfer Learning technique with learning rates of 0.1, 0.01, and 0.001. The VGG Net-19 architecture was found to have a validation accuracy of 92.6% for a learning rate of 0.001.
  • Advancing Corporate Finance: A Multigranularity Approach to Bankruptcy Prediction
    A Suresh, Durairaj K, B Anandan, K. D. Mohana Sundaram, B. Ravi Babu, Agan Prabu S
    2025 IEEE International Conference on Advanced Computing Technologies Icact 2025, 2025
    The global financial scene in 2025 is showing a rise in company failures, hence stressing the vital necessity of improved prediction technologies in corporate finance in a time characterised by ongoing inflationary pressures, supply chain interruptions, and geopolitical uncertainty. Designed to revolutionise bankruptcy prediction by combining multiresolution feature analysis with a strong Gradient Boosting Machine framework, this paper presents FIN-MGBoost, a revolutionary Financial Multi-Granularity Boosting Network. FIN-MGBoost achieves an extraordinary accuracy of 99.64% by using a thorough dataset of 6,819 Taiwanese companies from the Taiwan Economic Journal and exploiting its creative Multi-Granularity Feature Fusion (MGFF) module to extract deep insights from both macro-level financial indicators and micro-level anomalies. This model offers unmatched interpretability and scalability, hence overcoming the constraints of conventional statistical methods and previous machine learning techniques. It is improved by a near-perfect ROC curve (AUC <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$=\mathbf{0. 9 9}$</tex>) and low misclassification rates. By means of actionable information, FIN-MGBoost empowers a wide spectrum of stakeholders-creditors, financial analysts, auditors, and risk managers-to make strategic decisions and reduce risk in an economy growing more unstable.
  • Aero Dispense: Contactless QR Medicine Delivery
    Ratnakamala Petla, R L Krupakaran, Harathi Nimmala, K. D. Mohana Sundaram, K Elangovan, S. Kannapan
    2025 International Conference on Intelligent and Secure Engineering Solutions Cises 2025, 2025
    A QR code enabled smart medicine dispatcher (SMD) is designed to improve rural healthcare accessibility via a simplified, automated system of medication dispensing. The system consists of a mobile application with a QR code reader, enabling patients to verify and choose prescribed medicine. A microcontroller ESP32 processes the data from the QR code and operates the dispensing system with three servo motors for proper and safe distribution of the medicine. An LCD display provides immediate feedback, while a buzzer gives notifications to users in the event of successful medicine dispatch. The system is integrated with a telemedicine platform, whereby patients start consultations by scanning a QR code placed on the dispensing machine which leads to a phone call to the doctor. Over with phone conversation the physicians create electronic prescriptions as digital codes in the form of QR codes and these are dispatched to the patients. Once scanned, the smart medicine dispatcher authenticates the prescription and dispenses the prescribed medicine. IoT integration facilitates smooth communication and remote monitoring, minimizing human intervention and guaranteeing error-free medicine dispensation. A stable power supply ensures uninterrupted operation in distant areas, rendering the system a feasible solution for enhancing rural healthcare accessibility.
  • MIL-BOT Sentinel: Integrating Real-Time Spying and Remote Bomb Diffusion in Battlefield Environments
    B. Ravibabu, K. D. Mohana Sundaram, S. V. Rajesh Kumar, P. M. Vijayan, S. Venkatakiran, Malli Rajeswara Rao
    2025 IEEE International Conference on Intelligent Signal Processing and Effective Communication Technologies Inspect 2025, 2025
    This project presents a prototype of a batterypowered military spying and bomb disposal robot designed to operate in hazardous environments with enhanced security and remote-control capabilities. The robot is equipped with a Bluetooth module (HC-05) for manual control via the Arduino Bluetooth Controller App, enabling real-time buzzer alerts. For surveillance, a 360° Wi-Fi camera (CP Plus E24A) with two-way communication, auto night vision mode, siren alarm, and video storage is integrated, accessible via the Ezykam+App. A nightvision wireless camera further enhances live monitoring in lowlight conditions. The LCD display provides real-time movement status, indicating directions such as forward, backward, right, left, and stop. The robot’s mobility and operational control rely on two motor drivers: Motor Driver 1: Controls the robotic arm, including rotation and shoulder movement, ensuring precision in bomb disposal tasks. Motor Driver 2: Manages the robot’s wheels, enabling efficient terrain navigation. A metal detector and buzzer are integrated to detect and alert operators about metal-based explosives. The system is powered by three 3.7 V Li ion batteries for the robot and an additional $\mathbf{3. 7} \mathbf{V}$ Li-ion battery for the camera. This multi-functional robot combines mobility, surveillance, reconnaissance, and explosive handling capabilities, making it suitable for high-risk military operations in confined or hazardous areas.
  • Hybrid Filtering-Based Product Recommendation System Integrating GRU and BFGS Optimization
    A. Suresh, R. G. Kumar, D. Nagaraju, K D Mohana Sundaram, B Anandan
    IEEE International Conference on Electronic Systems and Intelligent Computing Icesic 2024 Proceedings, 2024
    It is more crucial than ever to handle crucial problems including data sparsity, cold-start problems, and the requirement for extremely precise forecasts in today’s ever changing recommendation system field. This work presents a hybrid filtering product recommendation system that integrates Gated Recurrent Units (GRU) with the Broyden-Fletcher-Goldfarb-Shanno (BFGS) optimization method to significantly enhance recommendation accuracy and performance. Standard techniques such content-based filtering and collaborative filtering can fail to offer reliable recommendations for large and dynamic datasets. Our hybrid model uses BFGS optimization to increase model accuracy and efficiency while utilizing GRU-based collaborative filtering to capture sequential user-item interactions. The GRU-BFGS model outperforms matrix factorization (MF) techniques in numerous areas, achieving an astounding accuracy of up to 92% on the Amazon customer review dataset. It also exhibits excellent recall, precision, and notable decreases in error metrics (MAE, MSE, and RMSE), which makes it a very successful method for providing individualized, accurate suggestions.
  • An efficient fruit quality monitoring and classification using convolutional neural network and fuzzy system
    K.D. Mohana Sundaram, T. Shankar, N. Sudhakar Reddy
    International Journal of Engineering Systems Modelling and Simulation, 2023
    Fruit quality monitoring in agro industries is carried out by people who may deviate from their responsibility due to tiredness, illness, or personal reasons. So, an automatic quality assessment system is proposed based on convolutional neural network (CNN) and Mamdani fuzzy logic that estimate quality of a Persian Lemon. The proposed CNN was trained with the transfer learning method and the results obtained were compared with previous works. The proposed CNN achieved 94.79% accuracy in the validation process which is 13% higher than the existing architecture. The proposed fuzzy logic classified each lemon in three ranges based on rules customised for the estimation of fruit quality standards.
  • RETRACTION:A novel fuzzy pooling based modified ThinNet architecture for lemon fruit classification
    K.D. Mohana Sundaram, T. Shankar, N. Sudhakar Reddy
    Journal of Intelligent and Fuzzy Systems, 2022
    Computer vision functions like object detection, image segmentation, and image classification were recently getting advance due to Convolutional Neural Networks (CNNs). In the food and agricultural industries, image classification plays a critical role in quality control. CNNs are made up of layers that alternate between convolutional, nonlinearity, and feature pooling. In this article, we proposed Fuzzy Pooling, a novel pooling approach that works based on fuzzy logic, that can increase the accuracy of the CNNs by replacing the conventional pooling layer. This proposed Fuzzy Pooling was put to the test with CIFAR-10 and SVHN data sets on single layer CNN, and it outperformed previous pooling strategies by achieving 92% and 97% classification accuracy. This proposed Fuzzy Pooling layer was replaced the Max Pooling layer in the ThinNet architecture, and it was trained using the back propagation method. It was demonstrated experimentally on the Lemon fruit data set to classify the fruits into three categories such as Good, Medium, and Poor. In order to classify the lemon fruit into three categories, the 1000 Fully Connected layer in ThinNet architecture was replaced with three Fully Connected layers. The Modified ThinNet architecture called ThinNet_FP was trained with a learning rate of 0.001 and achieved 97% accuracy in classifying the images and outperformed previous CNN architectures when trained on the same data set.